PROJECT SUMMARY Bipolar disorder (BD) is a devastating neuropsychiatric illness that affects 2-5% of youth and causes morbidity, functional impairment, and suicide. Prodromal manic symptoms without manic episodes usually emerge before BD types I and II (BD-I/II) develop, but less than 60% of youth with manic symptoms will develop BD-I/II. The uncertainty of diagnosis and illness progression results in potentially detrimental interventions and 7-10 years delay appropriate treatments. It is thus imperative that objective biomarkers of risk for conversion to BD-I/II are identified and tested in youth before the peak onset of illness. Given that neural measures of structure and function associated with emotion and reward processing, in combination with clinical and behavior measures, can improve prediction of psychiatric outcomes in youth, this project will investigate brain-behavior relations in the most severely ill youth during inpatient stays and aims to build a predictive model of BD. We aim to use two distinct analytic models to test our hypotheses. First a general linear model (GLM) with a machine learning (ML) model of regularized regression with cross validation and second a whole brain ML pattern recognition model. We will first identify neural and behavioral markers of BD-I/II in circuitry associated with emotion and reward processing. We hypothesize that decreased activity and connectivity in prefrontal, amygdala, and striatal regions and behavioral measures showing less sleep, lower activity, and poorer mood and cognition will distinguish BD- I/II from clinically matched youth without mania and healthy. Next, we will identify using ML a whole brain neural classifier of BD-I/II relative to clinically matched inpatients without mania. Aim 2 is to, after two years, identify and quantify the neural and behavioral measures that predict conversion to BD-I/II, and to test individual conversion in an independent group of high symptomatic risk adolescents. Aim 3 is to identify brain-behavior associations for app development. Training samples include mid-/post- pubertal adolescents aged 13-17 years recruited from the nation?s only specialized inpatient unit for adolescents with BD and the general adolescent unit at our hospital; 70 well-characterized adolescents with BD-I/II, a clinically matched group of 70 inpatient youth without mania. Testing sample is an independent group of 180 adolescents with manic symptoms without BD-I/II. 60 healthy controls will be recruited. The project includes emotion and reward processing neural function and structure, clinical and behavioral measures including sleep and activity with actigraphy, computerized cognitive measures, and self-reports during inpatient evaluation and for two weeks post discharge. At two-year follow up, clinical assessments will confirm diagnoses. This is the first study to employ a multimodal assessment of behavior and mood symptoms combined with multimodal imaging methods to comprehensively assess disease-specific abnormalities and prediction of BD-I/II. Findings from this study may identify biological and behavioral markers of conversion to BD-I/II in adolescents and may contribute developing disease-specific risk calculators, low-cost biosensors for mobile applications, and novel targets of intervention.